Error budget - Investigate mirror/pull API

Investigate mirror/pull API This is a follow up issue based on the work work scalability team had done in gitlab-com/gl-infra/scalability#1142

We want to reduce Pipeline authoring error budget and we discover the following

POST /api/:version/projects/:id/mirror/pull is the second highest contributor. This endpoint is fairly distributed across all organizations. There is no dominant. It means that a wide range of users are getting slow responses. The chart below showing that the 95th percentile of this API endpoint is 5.596 seconds.

we would like to investigate why we are getting a slow response rate from this API, more information in the related issue